Thanks to all who replied, on- and off-list.
At 2:32 PM -0400 6/1/01, Francois Lachance wrote:
>Walter, Peter, and other interested parties,
>Would the <ref> element with can take multiple values for its "target"
>attribute would do the trick? with perhaps the "corresp" attribute of the
>elements belonging to the pointer tag set to connect <div> and <ref>?
This might work syntactically, but I need the links to go the other
way: I want to say "this <div1> covers these GIS areas" rather than
saying "this GIS area is covered by these <div1>s. I think what I'm
going to end up doing is to use a feature library defining all values
applicable to a particular notebook, and use the ana attribute (since
it's IDREFS) to link the text to the data elements:
<f id="T003NR004W" name="TownshipRange"><str>Township 3 North,
Range 4 West</str></f>
<f id="T003NR005W" name="TownshipRange"><str>Township 3 North,
Range 5 West</str></f>
<f id="T003NR006W" name="TownshipRange"><str>Township 3 North,
Range 6 West</str></f>
<f id="WI" name="State"><sym value="Wisconsin"></f>
<f id="PM4" name="PrincipalMeridian"><str>4th Principal Meridian</str></f>
<f id="L1" name="SectionLine"><nbr value="1"></f>
<f id="L2" name="SectionLine"><nbr value="2"></f>
<f id="L3" name="SectionLine"><nbr value="3"></f>
<div1 type="line survey" ana="T003NR005W PM4 WI L2">...</div1>
One thing I'm not sure about is whether it's semantically kosher to
bypass <fs> entirely, and have the ana attribute point directly to
features in the <fLib>. It's certainly more efficient for this
application to do it this way, since the "feature structure" is
exactly the set of features referenced in the ana attribute.
I received a good suggestion to encode the document-level data
elements in a <creation> element in the header. That would probably
look like this:
<rs ana="T003NR005W T003NR005W T003NR006W PM4 WI">Township 3
North, Ranges 4-6 West, 4th Principal Meridian (Wisconsin)</rs>
Does this make sense? I like the fact that it lets me express the
data elements and relationships explicitly. It seems odd, though,
that <fLib> and <fsLib> can only occur within <text> -- as abstract
collections of values, it seems you'd want them in the <teiHeader>.
Not a problem, just an observation.